Article Text
Abstract
Objective To describe characteristics and outcomes of patients hospitalised for injuries occurring in industrial settings during a 1-year period.
Methods A retrospective analysis of hospital admissions in the USA in 2006 using the Nationwide Inpatient Sample was performed. The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9 CM) code E849.3 (industrial place and premises) was used to identify work-related injury admissions.
Results A total of 5826 patients were hospitalised with injuries sustained in industrial settings (weighted, 28 354 patients). The mean age was 42.9 years (82% were men). They were 48% Caucasian, 19% Hispanic and 6% African–American. The majority were admitted from the Emergency Department (72%). Further the majority of admissions were discharged home (79%; 9% with home healthcare) and 10.7% were transferred to another facility. The mean length of stay was 4.5 days (range 0–109 days). Mean total charges per admission was US$32 254 (median US$18 364, 90th percentile US$66 607). Common diagnoses included: orthopaedic injuries (including amputations) to: finger/hand (20.9%), foot/ankle (8.2%), leg (10.2%) and spine (8.4%); infection (10.8%), pulmonary diagnosis (6.6%), soft tissue injuries (3.6%) and burns to <10% of the body (3.6%). Comorbidities included hypertension (17.0%) and diabetes mellitus (6.3%). Most common procedures performed included fracture reduction (17.6%), blood transfusions (3.1%) and spinal surgery (3%). A total of 194 (0.7%) patients died in the hospital.
Conclusions Injuries in industrial settings result in significant healthcare usage, morbidity and mortality on an annual basis in the USA. These admission levels facilitate development of targeted strategies to optimise the quality and economics of care for injuries in industrial settings.
- Hand injury
- occupational injury
- case-crossover study
- epidemiology
- workplace
- multiple injury
- public health
- occupational
- falls
- elderly
- surveillance
Statistics from Altmetric.com
- Hand injury
- occupational injury
- case-crossover study
- epidemiology
- workplace
- multiple injury
- public health
- occupational
- falls
- elderly
- surveillance
Introduction
Costs associated with occupational injuries constitute a large proportion of total costs of medical care in the USA.1–9 Workplace injuries are associated with approximately US$140 billion annual costs including medical services (US$17 billion), productivity losses (US$60 billion), insurance (US$5 billion) and lost quality of life (US$62 billion).2 Non-economic effects on the quality of life including physical and psychological functioning as well as diminished well-being, that is, associated with occupational injures are more difficult to quantify.10 Worldwide, over 120 million occupational injuries are estimated to occur annually resulting in significant economic burden of up to several percent of the gross national product for some countries.3 ,9 ,11–13
Detailed information on the costs associated with occupational injuries reflects the incidence and severity of injuries occurring in the US and may be useful to policy makers to assist in determining appropriate and judicious resource allocation.14 Detailed cost information may also be useful for describing the range of injuries, conducting comparisons among different injury types, evaluating contribution to national healthcare costs and conducting cost–benefit analyses of implemented safety measures.2 ,9 ,14 ,15
Significant variations in injury and illness cost are expected across states due to differences in the composition of industries.16 ,17 An analysis of the Bureau of Labour Statistics data found that southern and western states had higher costs per worker than other states.17 Insurance coverage, admission type and the industry sector in which the injured workers were employed have been found to be significantly associated with treatment costs.16 ,18 ,19
Knowledge of the national data can be useful in determining practice guidelines and benchmarking.20 The objective of this study was to evaluate national trends of injuries occurring in industrial settings over 1 year, focusing on patient characteristics, costs of hospital care and outcomes.
Methods
The Nationwide Inpatient Sample (NIS) was used for this study. The NIS was developed as part of the Healthcare Cost and Utilisation Project, a Federal-State-Industry partnership sponsored by the Agency for Healthcare Research and Quality. Data from 1044 hospitals in 38 States or approximately 8 million hospital stays are included in NIS, approximating a 20% stratified sample of US community hospitals.21
The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9 CM) code E849.3 (industrial place and premises) was used to identify admissions of interest. This code includes: building under construction, dockyard, dry dock, factory building, premises, garage (place of work), industrial yard, loading platform (factory) (store), plant (industrial), railway yard, shop (place of work), warehouse and workhouse. Records that contained e-code E849 in any of the e-codes associated with each hospital admission were included in the analyses. We used total hospital charges as a surrogate for resource usage.22 ,23
Patient and hospital-specific variables were evaluated to determine those associated with increased resource usage. A full list of patient and hospital characteristics is available in the NIS dataset description.21 Briefly, the pertinent database variables included: admission day of week or weekend, admission month, admission source, admission type, age at admission, number of chronic conditions, diagnosis information, discharge quarter, discharge year, disposition of patient, external causes of injury and poisoning, gender of patient, length of stay, location of the patient, median household income for the patient's ZIP code, payer information, procedure information, race of patient, total charges, hospital location, hospital teaching status, hospital region, discharge weights, stratum and unique identifier.
Variables of interest included hospital location, admission source (emergency, outside hospital, routine), patient disposition and teaching/academic status. To have a teaching designation, a hospital is required to have: a residency programme approved by the American Medical Association, membership in the Council of Teaching Hospitals, or a ratio of full-time equivalent interns and residents to beds of at least 0.25.21
Statistical analysis
Basic descriptive weighted analyses were performed in SAS software (SAS, Cary, North Carolina, USA) using procedures that provided estimates of nationwide population statistics based on the NIS database. For example, PROC SURVEYFREQ, PROC SURVEYMEANS and PROC SURVEYREG were all used to calculate statistics based on the sampling design of the NIS dataset. Simple linear regression analyses were conducted to determine the significant predictors of increased resource usage as estimated by total charges. Total charges variable was considered an outcome and univariate analyses were conducted to determine the variables that were significant predictors of one unit increase in total charges. An α level of 0.05 was the level of significance and two-sided tests were used. Patient and hospital characteristics that were significant in the univariate analyses were evaluated in the multivariate model. Age and length of stay were included because of their clinical importance. Hospital location was selected as the only variable to indicate geography to avoid collinearity with patient location and hospital region.
This study was approved by the Office of Regulatory Affairs at the University of Nebraska Medical Center.
Results
There were 5826 records corresponding to a weighted total of 28 354 patients hospitalised with injuries sustained in industrial settings in the USA in 2006 (ICD-9 CM code E849.3) (table 1).
The majority of patients in this sample were men (82%). The average total charges per admission was higher for male than for female patients; US$33 069 (SE US$1965) compared to US$29 070 (SE US$1883), respectively (p value 0.01). The mean age was 42.9 years (SE 0.4). Caucasians constituted 48% of the patients, followed by 19% Hispanic and 6% African-American. There was no significant difference in total charges based on race.
The majority of admissions were from the Emergency Department (72%) and non-elective (87%). Trauma admissions accounted for 837 records (3%) and were associated with increased total charges compared to emergency admissions (p value <0.0001). The mean length of stay was 4.5 days (SE 0.2) per admission. Upon discharge, 22 464 (79%) patients were discharged home, 2405 (8.5%) patients were discharged with home healthcare and 3031 (10.7%) patients were transferred to another facility. There were 194 fatalities in this sample (0.7%). On average, fatalities were associated with the highest cost per admission, US$126 450 (SE US$22082) compared to US$24 689 (SE US$1198) for routine discharge (p value <0.0001).
The mean total charge per admission was US$32 254 (median US$18 364, 90th percentile US$66 607). The primary payer was other in 16 997 (60%) admissions, private insurance in 5325 (19%) admissions, Medicare/Medicaid in 3046 (11%) admissions and self-pay in 2766 (10%) admissions. Other primary payers included: Worker's Compensation, The Civilian Health and Medical Program of the Uniformed Services (in the USA) (CHAMPUS), The Civilian Health and Medical Program of the Department of Veterans Affairs (CHAMPVA), Title V and other government programmes. There was significant difference in charges based on payer with Medicare having the highest charges per admission (US$36 240, SE US$2858, p value 0.0006).
There was significant difference in charges based on hospital region or hospital location. The majority of hospitals were in urban locations (24933, 88%) and were associated with higher total charges per admission (US$24 304, SE US$1902) than hospitals in rural areas (US$17 615, SE US$2045) (p value <0.0001). In addition, hospitals located in the West had highest total charges compared to other regions (p value <0.0001).
Information on the number of admissions for injuries in industrial settings in each state and their respective mean charges per admission are presented in figure 1. The highest number of records was from California (6156, SD 1086), followed by New York (3794, SD 740.2), Florida (2028, SD 669.2), Texas (1585, SD 600.2) and Missouri (1313, SD 435.3). Records from these five states constituted 52% of all records in the NIS sample. The highest mean total charges per admission were for Nebraska (US$74 162, SE US$1581.4, 294 records), Florida (US$49 788, SE US$8635.9, 2029 records), California (US$47 502, SE US$5452.2, 6156 records), Nevada (US$37 840, SE US$4353.2, 514 records) and Colorado (US$37 598, SE US$6150.3, 1020 records). Kansas had the lowest mean charge (US$11 079, SE US$2521.0) for 110 admissions followed by Maryland (US$11 241, SE US$295.0) for 901 admissions.
Multivariate analysis included gender, admission type, discharge status, primary payer, patient location, hospital location (urban/rural) and hospital region. Age and length of stay were included because of their clinical importance. Hospital location was selected as the only variable to indicate geography to avoid collinearity with patient location and hospital region. The multivariate model was based on 3357 records (59%) with R2 of 0.54. With the exception of admission type all variables remained significant predictors of increased total charges in the multivariate model. (table 2) After adjusting for important hospital and patient characteristics, on average, the total charge for the 60 years and over age group was US$5721 less than the 20–29 year age group (p=0.032), but none of the other age groups differed significantly from the 20–29 age group. Total charges for Asian/Pacific Islanders were on average US$13 000 higher than Caucasians and Native Americans were US$5590 higher than Caucasians (p=0.031 for both). Total charges were on average US$5711 less in rural hospitals than in urban hospitals (p=0.001). Table 2 contains additional estimated differences for length of stay, discharge status and primary payer, for the multivariate model.
Procedure and/or diagnostic codes were reviewed to determine the most common diagnoses and procedures performed in this sample. A total of 52% of patients had orthopaedic injuries (including amputations) to: finger/hand (20.9%), foot/ankle (8.2%), leg (10.2%) and spine (8.4%). Infections were found in 10.8% of patients, 6.6% included a pulmonary diagnosis, 3.6% had soft tissue injuries and 3.6% had burns <10% of their body. Of all patients, 22.2% used tobacco and 1.2% had a history of alcohol misuse. In addition, 17.0% of patients were hypertensive and 6.3% had diabetes mellitus. The most common procedures included fracture reduction (17.6%), blood transfusions (3.1%) and spinal surgery (3%). A total of 194 (0.7%) patients died in the hospital.
Discussion
This study evaluates and discusses the costs associated with injuries occurring in industrial settings in the US using a national database of hospital discharge records. A total of 28 354 patients were hospitalised with injuries sustained in industrial settings in the USA in 2006. Other studies of occupational injuries have relied on payer information or prospective data collection and their results may not be directly comparable to the results of our study. As expected for occupational injuries, the majority of patients in this sample were men with the average total charges per admission of US$33 069 compared to US$29 070 for female patients (p value 0.014).19 ,24 This may be partially explained by the types of occupations as male workers tend to be in more hazardous occupations and conditions than their female counterparts.25 As injury severity is associated with costs, it is expected that the gender differences in costs reflect injuries of higher severity sustained by male workers.12 ,26 ,27
The demographic characteristics of the groups of patients in this study were comparable to those reported in other studies of occupational injuries.24–27 The majority of patients in this sample had orthopaedic injuries to upper and lower extremities or spine. Diagnoses of infections, pulmonary diagnosis, soft tissue injuries and burns were also common. Consistently, the most common procedures included fracture reduction, blood transfusions and spinal surgery. With the exception of fatalities, we were unable to quantify injury severity using these data due to inherent limitations of the database.
The mean age of patients sustaining injuries in industrial settings and being admitted to a hospital was 42.9 years. The average age was comparable to average ages reported in other studies.18 ,19 Older workers are expected to incur higher costs possibly due to comorbidities and complications.25 ,28 There were no significant differences in total charges based on race; the majority of patients were Caucasian, followed by Hispanic and African–American. Others have found that compared to Caucasian non-Hispanic workers, the costs of non-fatal injuries of US construction workers were higher for African–American non-Hispanic workers25 and for Hispanic workers.29 This may be due to myriad of factors including more dangerous jobs, occupations of minority workers, and reluctance of this group to seek medical treatment for minor injuries.29 There may be contributing factors to economic disparities including societal, language and cultural29 although we did not find any disparities in this sample.
The length of stay reported in this study was similar to other such reported studies. For example, a study of worker compensation data found that on average, workers required 4.0 days of hospitalisation.18 Large variations in the length of stay associated with occupational injuries, however, have been reported.19 ,24
Comparable to other studies, admission type was significantly associated with total charges.26 ,28 The number of emergency admissions was much higher than the trauma admissions. As expected, the average cost per admission, however, was much higher for trauma than for emergency admissions. There were 194 fatalities (0.7%) in this sample, which were associated with over five times higher total charges compared to routine discharge patients. This was expected, because total charges reflect injury severity and fatalities were expected to be associated with injuries of the highest severity.
Primary payer was significantly associated with total charges (p=0.0006). Other and private insurance paid for the majority of admissions, although the cost per admission was the highest for admissions paid by Medicare. Medicare is a US governmental social insurance programme, which provides health insurance coverage for those aged 65 or older, or those who have permanent physical disability, congenital physical disability, or those who meet other special criteria like end-stage renal disease. Higher costs for workers' compensation admissions were previously reported with more procedures per stay administered for workers compensation cases compared to non-workers compensation patients, which may partially explain higher costs.18 Costs were significantly lower for admissions in the micropolitan and non-core areas and in rural hospitals. The teaching status of the hospital was not significantly associated with total charges.
Costs were significantly higher in the Western region. Of the states with the highest frequency of injuries occurring in industrial settings, Florida was associated with the highest (US$49 788) and Maryland with the lowest (US$11 241) mean charges per admission. Nebraska had the highest total charges per admission with a relatively low number of admissions. Large regional differences in the number of records and total charges found may reflect variations in demographics, industry and reporting. For example, demanding manufacturing jobs in the Midwest may employ older workers while service industry in the West may have a relatively younger worker population.30
The main limitation of this study was using an administrative national database that did not allow evaluation of an in-depth admission data. We used e-codes to determine whether an occupational injury had occurred. E-codes, however, may not be the best indicator for occupational injuries and the true number of admissions was likely to be underestimated using this approach. One systematic review found that accuracy of external cause coding in ICD-9 CM ranged from 64% to 85% depending on whether the exact code agreement or a broader group agreement was examined.31 A number of admissions may have been missed if e-code for industrial place or premise was not included in any of the four e-codes associated with that admission. It is also possible that a small number of injuries that occurred at an industrial places or premises were not work related. We expect the number of false negatives to be much larger than the number of false positives using this ascertainment method.31
Total charges rather than hospital costs are captured by the database and as such were used in our analyses. The validity of using total charges as a surrogate for total costs has been extensively detailed using this database and has been successfully used in other settings.22 In this database, only inpatient costs are included in total charges. Solely using total charges can introduce bias and may not be accurate in the absolute sense; however it is standardised and facilitates the comparisons, which had been performed herein.
While workers compensation data contain limitations,8 ,32 it would be preferable to estimate the cost associated with occupational injuries using workers compensation as primary or secondary payer.18 ,33 Due to the changes in NIS database coding, a workers compensation indicator was not available for the 2006 data. In addition, potential contributing factors to costs associated with occupational injuries such as nature and location of the job and injury severity could not be evaluated.16 More severe or multiple co-occurring injuries are expected to have a higher number of procedures that would increase associated costs.19 ,28
Finally, the quality of the large national database used in this study relies on accurate reporting from the participating institutions. Despite these limitations, the NIS database provided a unique perspective on hospitalisations for injuries occurring in industrial settings. The large number of records available for analyses allows detailed evaluation of demographic variables associated with occupational injuries. These admission level data can facilitate development of targeted strategies to optimise the quality and economics of care for injuries in industrial settings.
Conclusions
This study presented a national perspective on patients hospitalised for injuries occurring in industrial settings. There was significant variation in total charges among US states and micropolitan rural settings were associated with decreased charges. The results of this study have important implications with regards to the economics of occupational injuries. Cost data may be used to develop and implement policies on cost containment. Detailed study on a micro level is warranted to evaluate these variances in greater detail in order to better understand their impact on outcomes and resource usage (eg, specific outcomes for specific diseases, the precise amount of usage per diagnosis, etc) which are not possible on an aggregate, macro level. Furthermore, such normative data provide opportunities for benchmarking and goal setting in understanding prospective opportunities to improve the economic implications of occupational injuries in the USA. On a broader scale, country-to-country comparisons using such data will identify further opportunities for strategic interventions.
What is already known on the subject
-
Occupational injuries are associated with significant economic burden worldwide.
What this study adds
-
We describe the characteristics and outcomes of patients hospitalised for injuries occurring industrial settings over 1 year using a large national database (28 354 admissions).
-
Mean total charges per admission for injuries occurring in industrial settings was US$32 254 (median US$18 364, 90th percentile US$66 607).
-
A total of 52% of patients had orthopaedic injuries to the upper and lower extremities and spine. The most common procedures were fracture reduction (17.6%), blood transfusions (3.1%) and spinal surgery (3%).
-
There exists significant variation in total charges among US states.
References
Footnotes
-
Competing interests None.
-
Ethics approval Ethics approval was provided by University of Nebraska Medical Center Institutional Review Board.
-
Provenance and peer review Not commissioned; externally peer reviewed.